Automic Workload Automation

Expand all | Collapse all

Runtime Monitoring - Maximum runtime (MRT) lower than 1 minute is not reliable detected

  • 1.  Runtime Monitoring - Maximum runtime (MRT) lower than 1 minute is not reliable detected

    Posted 03-21-2017 08:03 AM

    The Automation Engine is not a real time engine. All time based actions are triggered by a so called “Timer” transaction. This transaction is executed every 20 seconds. (In realty it is a little more complex, however knowing this is sufficient to understand the behavior explained below.)

    Example:

    A UNIX job executing the os command “sleep 30” will have a runtime of approximately 30 seconds:

    atmrp5c40dra.pnghttps://us.v-cdn.net/5019921/uploads/editor/wg/atmrp5c40dra.png" width="664">

    Using maximum runtime (MRT) of 15 seconds, else cancel, should abort the job:

    6q1431sz8i1x.pnghttps://us.v-cdn.net/5019921/uploads/editor/0w/6q1431sz8i1x.png" width="715">

    However it may happen, that the job is not canceled:

    id6dunctxu0m.png

     

    Let’s think about 2 scenarios:

    1. First Timer take place 5 seconds after job start – MRT still ok.
      Next Timer will be 25 seconds (= 5 + 20) after jobs start. Job is still active and will be canceled, because MRT is violated.

    2. First Timer take place 12 seconds after job start – MRT still ok.
      Next Timer will be 32 seconds (= 12 + 20) after jobs start. Job has finished in the meantime (2 seconds before), so it cannot be canceled, even MRT is violated.

    Note:In addition the timing might also be influenced by the current load of the system.

    Knowing this it is understandable, that maximum runtime setting below 1 minute do not make sense.




  • 2.  Runtime Monitoring - Maximum runtime (MRT) lower than 1 minute is not reliable detected

    Posted 08-11-2017 05:47 AM
    I wish I'd found this article earlier.  I couldn't get this feature to work but within seconds of reading this I'd got it working.

    BIG thank you